Practical Deep Learning for Coders from fast.ai is a free deep learning course built on PyTorch, fastai and Hugging Face.
Courses
14 courses
Free to take
3 courses
Curated by Jake BolandChecked against the issuers' and providers' pages, Oct 12, 2026 to Oct 26, 2026How I verify
For a short paid option with a certificate, NVIDIA's Getting Started With Deep Learning takes 8 hours.
What do you need before a deep learning course?
Python and some machine learning basics, since most of these courses are intermediate level. If you are not there yet, the Machine Learning Specialization from Stanford Online and DeepLearning.AI is beginner level and paid, given as "2 months at 10 hours a week". Harvard's free CS50's Introduction to Artificial Intelligence with Python reaches neural networks in Python projects and comes with a free certificate. If you already code and want one course, fast.ai is the free pick; NVIDIA's is the one to choose if a certificate matters to you.
Free deep learning course or paid one?
The difference that matters is the certificate. fast.ai's Practical Deep Learning for Coders covers vision, NLP, tabular data and deployment, and fast.ai describes it as "9 lessons, and each lesson is around 90 minutes long". It doesn't state whether you get a certificate. NVIDIA's Getting Started With Deep Learning is a paid 8-hour introduction with a DLI certificate. For language models in particular, Hugging Face's LLM Course is free, though it gives no certificate.
Which longer programs include deep learning?
Two IBM programs on Coursera, both paid subscriptions. The IBM AI Engineering Professional Certificate teaches deep learning with Keras, PyTorch and TensorFlow inside a wider machine learning program, given as "4 months at 10 hours a week". The IBM Generative AI Engineering Professional Certificate starts at beginner level and goes through deep learning to transformers and fine-tuning, given as "6 months at 6 hours a week".
What if you run the hardware rather than the models?
NVIDIA's AI Infrastructure and Operations Fundamentals is for IT, network and data center staff. It covers GPU computing, storage and cluster scheduling, costs $50 and takes 7 hours 25 minutes. None of these deep learning courses ends in a certification exam, so the proof you get is a course certificate, where one is offered.
More deep learning courses
Recommended order: courses mainly about Deep learning first, then open now, wide audience, beginner level and lower cost. The 8 courses shown above are not repeated here unless you filter or search. How we order courses.